Conjoint analysis is one of the most powerful methodologies in market research. It tells you what people actually value when they are forced to make tradeoffs. Not what they say matters, but what drives their real decisions when they have to choose between competing options. Pricing, features, packaging, brand, messaging: conjoint quantifies the relative importance of each.
It is also one of the most inaccessible tools in the industry. A full-service conjoint project from a research firm typically costs $80,000 to $250,000. Even running one in-house requires $10,000+ in software licensing, $5,000 to $150,000 in panel recruitment, a specialized analyst, and 8 to 16 weeks of end-to-end timeline. The teams that would benefit most from conjoint, product teams validating feature priorities, brand teams testing positioning, smaller research orgs running pricing studies, are usually the ones who can least afford it.
Simsurveys now offers synthetic conjoint analysis. The same methodology, the same statistical rigor, at a fraction of the cost and timeline. This post explains what that means, how it works, and why it matters.
What Is Conjoint Analysis?
Conjoint analysis is an experimental methodology that measures how people value different attributes of a product or service. Rather than asking respondents to rate individual features in isolation, conjoint forces them to make tradeoffs between complete product profiles, just like they would in a real purchase decision.
The most widely used form is choice-based conjoint (CBC). Respondents see sets of 3 to 5 product profiles, each defined by a different combination of attribute levels (for example: price at $9.99 vs. $14.99, brand A vs. brand B, feature included vs. not included). They pick the one they would choose. This is repeated across 12 to 20 choice tasks per respondent, with attribute levels varied according to a statistically optimized experimental design.
From these choices, the model estimates part-worth utilities for every attribute level: numerical scores that represent how much each feature contributes to (or detracts from) a respondent's preference. Higher utility means stronger preference. The gap between the highest and lowest utility within an attribute tells you how important that attribute is relative to the others.
The result is a precise, quantitative map of what drives customer decisions. Not opinions, not ratings, but revealed preferences derived from actual choice behavior.
Why Conjoint Is So Expensive
The cost of conjoint analysis comes from the cumulative burden of its requirements. Each step demands specialized resources, and the costs compound quickly.
Software licensing. The industry-standard platform, Sawtooth Software's Lighthouse Studio, starts at roughly $16,500 per year for a single user. Their cloud-based Discover product starts around $7,500 per year. Alternatives like Conjointly and 1000minds are cheaper but still run $3,000 to $25,000 annually depending on the plan.
Experimental design. A conjoint study requires careful design: selecting the right attributes and levels, building a statistically efficient experimental plan, and programming the survey instrument. This typically requires a researcher with specific conjoint expertise. Most teams do not have one on staff.
Panel recruitment. You need real respondents to complete the choice tasks. For general population studies, panel recruitment runs $15 to $80 per complete, putting a typical n=300 to n=500 study at $5,000 to $25,000. For specialty audiences like physicians or C-suite executives, costs jump to $150 to $500+ per complete, and a single study can easily exceed $100,000 in panel costs alone.
Fielding time. Recruiting respondents, fielding the survey, and collecting enough completes typically takes 2 to 4 weeks for general population, longer for niche audiences. Add 2 to 4 weeks on either side for design and analysis, and the full project timeline stretches to 8 to 16 weeks.
Analysis and estimation. Once data is collected, utilities must be estimated using hierarchical Bayes multinomial logit (HB-MNL) models, the gold-standard estimation method. This requires someone who understands Bayesian estimation, MCMC convergence diagnostics, and how to build market simulators from the resulting utility vectors. That is a specialized skill set.
Add it all up and a minimum viable conjoint project runs $30,000 to $50,000 with in-house capability. Through a full-service agency, $80,000 to $250,000 is typical. For pharma or healthcare conjoint studies with specialty HCP panels, budgets regularly exceed $200,000.
What Is Synthetic Conjoint?
Synthetic conjoint analysis uses AI-generated respondents to complete conjoint choice tasks instead of recruiting a live panel. The respondents are produced by domain-specific models trained on real population data, generating choice behavior that is statistically consistent with how real people make tradeoffs.
The methodology is identical to traditional conjoint. The difference is the source of the respondents.
You still define your attributes and levels. You still get a statistically optimized experimental design. The synthetic respondents still complete choice tasks, selecting their preferred option from each choice set. HB-MNL estimation still runs on the choice data to extract individual-level part-worth utilities. And you still get the same deliverables: utility scores, attribute importance weights, preference share simulations, and willingness-to-pay estimates.
What changes is the cost, the timeline, and the accessibility. No panel recruitment. No fielding delays. No six-figure budget. Results in minutes instead of months.
How It Works
The synthetic conjoint workflow follows four steps.
Step 1: Define the conjoint design. You specify the attributes you want to test (for example: price, brand, key features, packaging format) and the levels within each attribute (for example: $9.99 / $14.99 / $19.99). Simsurveys generates a statistically efficient experimental design, the set of choice tasks that maximizes the information extracted from each respondent's choices.
Step 2: Generate synthetic respondents. Our AI models generate a representative sample of synthetic respondents matched to your target population. Each respondent carries a complete demographic and psychographic profile. They complete the full set of conjoint choice tasks, making selections that reflect learned preference patterns from real population data.
Step 3: Run HB-MNL estimation. Once the choice data is collected, we run full hierarchical Bayes multinomial logit estimation. HB-MNL is the same method used by Sawtooth Software and every major conjoint research firm. It uses MCMC sampling (specifically Gibbs sampling) to estimate individual-level part-worth utilities for every respondent, not just population averages. Each respondent gets their own utility vector, meaning you can analyze preferences at the individual level, not just in aggregate.
Step 4: Deliver the analysis package. You receive a complete conjoint deliverable:
- Part-worth utilities for every attribute level, at both the individual and aggregate level
- Attribute importance weights showing which attributes drive the most decision impact
- Preference share simulations that predict market share for any hypothetical product configuration
- Willingness-to-pay estimates derived from the price utility curve
- Segment-level analysis breaking down preferences by demographic or attitudinal subgroups
- Raw data export in CSV or SPSS format for your own analysis
Why HB-MNL Estimation Matters
The estimation method is what separates rigorous conjoint from a glorified survey. HB-MNL (hierarchical Bayes multinomial logit) is the gold standard for a reason.
The multinomial logit layer models the probability that a respondent chooses a given option as a function of the utilities of its attributes. The hierarchical Bayes layer adds a critical capability: it estimates utilities at the individual level, not just the population level. Each respondent gets their own unique set of part-worth utilities, drawn from a population-level distribution and refined by their own choice data.
This matters because aggregate utilities hide the most interesting insights. A population average might show that price and brand are equally important, but individual-level analysis might reveal two distinct segments: one that is highly price-sensitive and brand-indifferent, and another that will pay a premium for a trusted brand. HB-MNL surfaces those differences.
The estimation uses MCMC (Markov Chain Monte Carlo) sampling, specifically Gibbs sampling, to iteratively refine the utility estimates. This is computationally intensive but produces stable, reliable individual-level estimates even when each respondent only completes 12 to 15 choice tasks.
Simsurveys runs full HB-MNL estimation on every synthetic conjoint study. This is the same method that Sawtooth Software's CBC/HB module uses, and the same method that research firms charge six figures to run.
Use Cases
Pricing optimization. Conjoint is the most reliable method for understanding price sensitivity. By including price as an attribute, you can model exactly how much demand drops (or holds) at each price point, and how price interacts with other features. Synthetic conjoint makes this accessible for product launches, SKU rationalization, and competitive pricing scenarios without a $100K+ research commitment.
Feature prioritization. When your product roadmap has more candidate features than you can build, conjoint tells you which ones actually move the needle. Instead of relying on survey ratings where everything scores "important," conjoint forces tradeoffs and reveals the true hierarchy of value.
Packaging and bundling. What combination of features, services, and price points maximizes appeal? Conjoint's market simulator lets you test thousands of product configurations and predict which ones capture the most share. Synthetic conjoint lets you iterate on configurations rapidly without going back to field each time.
Competitive positioning. By including competitor products as profiles in the conjoint design, you can model how your offering stacks up against the competitive set. Preference share simulations show you where you win, where you lose, and what it would take to shift share.
Messaging and claims testing. Conjoint can test which claims or messages have the most impact on preference when positioned alongside product attributes. This moves messaging decisions from subjective judgment to quantified impact.
Cost and Timeline Comparison
The difference is not marginal. It is structural.
A traditional full-service conjoint project: $80,000 to $250,000, 8 to 16 weeks, requires specialized conjoint expertise, panel recruitment, and Sawtooth or equivalent software licensing.
A synthetic conjoint study through Simsurveys: a fraction of the cost, results in minutes, no panel recruitment, no software licensing, complete analysis package included.
This does not mean synthetic conjoint replaces every traditional conjoint study. For high-stakes final decisions where a board needs to see that real consumers were surveyed, traditional conjoint still has a role. But for the vast majority of use cases, where the goal is to understand preference structure, test configurations, and inform product decisions, synthetic conjoint delivers the same analytical output at radically different economics.
Getting Started
Synthetic conjoint analysis is available now through Simsurveys. You define your attributes and levels, select your target population, and receive a complete conjoint analysis package: utilities, importance weights, preference share simulations, and raw data export.
If you have run conjoint studies before, you will recognize every element of the output. If you have never been able to afford one, this is your chance to access the same methodology that the largest research firms charge six figures for.
Reach out for early access, a demo, validation studies, or any questions.